webpac2/trunk/run.pl

#!/usr/bin/perl -w

use strict;

use Cwd qw/abs_path/;
use File::Temp qw/tempdir/;
use Data::Dumper;
use lib './lib';

use WebPAC::Common 0.02;
use WebPAC::Lookup;
use WebPAC::Input::ISIS;
use WebPAC::Store 0.03;
use WebPAC::Normalize::XML;
use WebPAC::Output::TT;
use WebPAC::Output::Estraier 0.05;
use YAML qw/LoadFile/;
use LWP::Simple;

my $limit = shift @ARGV;

my $config = LoadFile('conf/config.yml');

print "config = ",Dumper($config);

die "no databases in config file!\n" unless ($config->{databases});

my $total_rows = 0;

while (my ($database, $db_config) = each %{ $config->{databases} }) {

        my $log = _new WebPAC::Common()->_get_logger();

        #
        # open Hyper Estraier database
        #

        my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration");
        $est_config->{database} = $database;

        $log->info("using HyperEstraier URL $est_config->{masterurl}");

        my $est = new WebPAC::Output::Estraier(
                %{ $est_config },
        );

        #
        # now WebPAC::Store
        #
        my $abs_path = abs_path($0);
        $abs_path =~ s#/[^/]*$#/#;

        my $db_path = $config->{webpac}->{db_path} . '/' . $database;

        $log->info("working on $database in $db_path");

        my $db = new WebPAC::Store(
                path => $db_path,
                database => $database,
                debug => 1,
        );


        #
        # now, iterate through input formats
        #

        my @inputs;
        if (ref($db_config->{input}) eq 'ARRAY') {
                @inputs = @{ $db_config->{input} };
        } elsif ($db_config->{input}) {
                push @inputs, $db_config->{input};
        } else {
                $log->info("database $database doesn't have inputs defined");
        }

        foreach my $input (@inputs) {

                my $type = lc($input->{type});

                die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');

                my $lookup = new WebPAC::Lookup(
                        lookup_file => $input->{lookup},
                );

                $log->info("working on input $input->{path} [$input->{type}]");

                my $isis = new WebPAC::Input::ISIS(
                        code_page => $config->{webpac}->{webpac_encoding},
                        limit_mfn => $input->{limit},
                        lookup => $lookup,
                );

                my $maxmfn = $isis->open(
                        filename => $input->{path},
                        code_page => $input->{encoding},        # database encoding
                );

                my $n = new WebPAC::Normalize::XML(
                #       filter => { 'foo' => sub { shift } },
                        db => $db,
                        lookup_regex => $lookup->regex,
                        lookup => $lookup,
                        prefix => $input->{name},
                );

                $n->open(
                        tag => $input->{normalize}->{tag},
                        xml_file => $input->{normalize}->{path},
                );

                for ( 0 ... $isis->size ) {

                        my $row = $isis->fetch || next;

                        my $mfn = $row->{'000'}->[0] || die "can't find MFN";

                        my $ds = $n->data_structure($row);

                        $est->add(
                                id => $input->{name} . "#" . $mfn,
                                ds => $ds,
                                type => $config->{hyperestraier}->{type},
                        );

                        $total_rows++;
                }

        };

        $log->info("$total_rows records indexed");

        #
        # add Hyper Estraier links to other databases
        #
        if (ref($db_config->{links}) eq 'ARRAY') {
                foreach my $link (@{ $db_config->{links} }) {
                        $log->info("adding link $database -> $link->{to} [$link->{credit}]");
                        $est->add_link(
                                from => $database,
                                to => $link->{to},
                                credit => $link->{credit},
                        );
                }
        }

}

1	dpavlin	74	#!/usr/bin/perl -w
2
3			use strict;
4
5			use Cwd qw/abs_path/;
6			use File::Temp qw/tempdir/;
7			use Data::Dumper;
8			use lib './lib';
9
10	dpavlin	255	use WebPAC::Common 0.02;
11	dpavlin	74	use WebPAC::Lookup;
12			use WebPAC::Input::ISIS;
13	dpavlin	209	use WebPAC::Store 0.03;
14	dpavlin	74	use WebPAC::Normalize::XML;
15			use WebPAC::Output::TT;
16	dpavlin	255	use WebPAC::Output::Estraier 0.05;
17	dpavlin	141	use YAML qw/LoadFile/;
18	dpavlin	210	use LWP::Simple;
19	dpavlin	74
20	dpavlin	76	my $limit = shift @ARGV;
21
22	dpavlin	141	my $config = LoadFile('conf/config.yml');
23
24			print "config = ",Dumper($config);
25
26	dpavlin	210	die "no databases in config file!\n" unless ($config->{databases});
27	dpavlin	141
28	dpavlin	213	my $total_rows = 0;
29
30	dpavlin	210	while (my ($database, $db_config) = each %{ $config->{databases} }) {
31	dpavlin	141
32	dpavlin	255	my $log = _new WebPAC::Common()->_get_logger();
33
34			#
35			# open Hyper Estraier database
36			#
37
38			my $est_config = $config->{hyperestraier} \|\| $log->logdie("can't find 'hyperestraier' part in confguration");
39			$est_config->{database} = $database;
40
41			$log->info("using HyperEstraier URL $est_config->{masterurl}");
42
43			my $est = new WebPAC::Output::Estraier(
44			%{ $est_config },
45			);
46
47			#
48			# now WebPAC::Store
49			#
50	dpavlin	210	my $abs_path = abs_path($0);
51			$abs_path =~ s#/[^/]*$#/#;
52	dpavlin	141
53	dpavlin	210	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
54	dpavlin	74
55	dpavlin	255	$log->info("working on $database in $db_path");
56
57	dpavlin	210	my $db = new WebPAC::Store(
58			path => $db_path,
59	dpavlin	217	database => $database,
60	dpavlin	221	debug => 1,
61	dpavlin	210	);
62	dpavlin	74
63	dpavlin	233
64	dpavlin	213	#
65			# now, iterate through input formats
66			#
67	dpavlin	74
68	dpavlin	213	my @inputs;
69			if (ref($db_config->{input}) eq 'ARRAY') {
70			@inputs = @{ $db_config->{input} };
71	dpavlin	255	} elsif ($db_config->{input}) {
72			push @inputs, $db_config->{input};
73	dpavlin	213	} else {
74	dpavlin	255	$log->info("database $database doesn't have inputs defined");
75	dpavlin	213	}
76	dpavlin	74
77	dpavlin	213	foreach my $input (@inputs) {
78	dpavlin	233
79			my $type = lc($input->{type});
80
81			die "I know only how to handle input type isis, not '$type'!\n" unless ($type eq 'isis');
82
83			my $lookup = new WebPAC::Lookup(
84			lookup_file => $input->{lookup},
85			);
86
87	dpavlin	213	$log->info("working on input $input->{path} [$input->{type}]");
88	dpavlin	74
89	dpavlin	213	my $isis = new WebPAC::Input::ISIS(
90			code_page => $config->{webpac}->{webpac_encoding},
91			limit_mfn => $input->{limit},
92	dpavlin	251	lookup => $lookup,
93	dpavlin	213	);
94	dpavlin	113
95	dpavlin	213	my $maxmfn = $isis->open(
96			filename => $input->{path},
97			code_page => $input->{encoding}, # database encoding
98			);
99	dpavlin	113
100	dpavlin	213	my $n = new WebPAC::Normalize::XML(
101			# filter => { 'foo' => sub { shift } },
102			db => $db,
103			lookup_regex => $lookup->regex,
104			lookup => $lookup,
105	dpavlin	221	prefix => $input->{name},
106	dpavlin	213	);
107	dpavlin	113
108	dpavlin	213	$n->open(
109			tag => $input->{normalize}->{tag},
110			xml_file => $input->{normalize}->{path},
111	dpavlin	210	);
112
113	dpavlin	213	for ( 0 ... $isis->size ) {
114	dpavlin	210
115	dpavlin	213	my $row = $isis->fetch \|\| next;
116
117			my $mfn = $row->{'000'}->[0] \|\| die "can't find MFN";
118
119			my $ds = $n->data_structure($row);
120
121			$est->add(
122			id => $input->{name} . "#" . $mfn,
123			ds => $ds,
124			type => $config->{hyperestraier}->{type},
125			);
126
127			$total_rows++;
128			}
129
130	dpavlin	210	};
131
132			$log->info("$total_rows records indexed");
133	dpavlin	255
134			#
135			# add Hyper Estraier links to other databases
136			#
137			if (ref($db_config->{links}) eq 'ARRAY') {
138			foreach my $link (@{ $db_config->{links} }) {
139			$log->info("adding link $database -> $link->{to} [$link->{credit}]");
140			$est->add_link(
141			from => $database,
142			to => $link->{to},
143			credit => $link->{credit},
144			);
145			}
146			}
147
148	dpavlin	210	}
149